[Bug] Custom API / LiteLLM provider saves successfully but models do not appear in chat model picker on Windows v0.37.0

[lhylvsea]

Environment:

• CodePilot v0.37.0
• Windows
• Provider type: Custom API (OpenAI-compatible)
• Base URL: http://127.0.0.1:4000
• Backend proxy: LiteLLM
• Model exposed by LiteLLM: nvidia-glm47

What I expected:

• After adding the provider, I should be able to select the provider/model in the conversation header.
• Docs say adding a provider includes selecting a default model, and providers can be switched in the conversation header.

What actually happened:

• The provider is saved and appears under "Connected Providers".
• LiteLLM /chat/completions works.
• LiteLLM /models returns nvidia-glm47.
• But the chat model dropdown only shows Claude Code models (Sonnet/Opus/Haiku).
• The Custom API edit dialog does not show any "default model" selection step.

Repro steps:

1. Run LiteLLM locally on http://127.0.0.1:4000
2. Verify /models returns nvidia-glm47
3. Add provider in CodePilot:
   • Name: litellm-local
   • Base URL: http://127.0.0.1:4000
   • API Key:
4. Save provider
5. Open a new chat and check the model picker

Additional verification:

• POST http://127.0.0.1:4000/chat/completions works
• GET http://127.0.0.1:4000/models works

[op7418]

Fixed in v0.38.1

This issue has been addressed with two fixes:

1. Model picker now synthesizes entries from role_models_json: If you set a default model name in the provider settings (e.g. nvidia-glm47), it will now appear in the chat model picker even if the provider has no pre-defined model catalog.

2. Doctor detects empty model lists: The Provider Doctor now warns when a provider has no models configured, with a clear message to set a default model name.

Please update to v0.38.1: https://github.com/op7418/CodePilot/releases

After updating:

1. Edit your LiteLLM provider in Settings → Providers
2. In the model name field (under Advanced Options), enter your model name (e.g. nvidia-glm47)
3. Save — the model should now appear in the chat picker

If it still doesn't work:

1. Go to Settings → Providers → Run Diagnostics
2. Click "Export Logs"
3. Attach the log file here